How to Build Your Own Face Blur Tool

Changsin Lee
5 min readJul 23, 2021

What does it take to build your own face blur tool? Why and how do you go about doing it?

· How to Customize De-identifying Faces using OpenCV
· 1. Detect Faces
· 2. Anonymize Faces
· a. Blurring in rectangles
· b. Blurring in ellipses
· c. Pixelating
· Conclusion
· Reference

How to Customize De-identifying Faces using OpenCV

DIY face blur tool

De-identification refers to the process of anonymizing data. With the enforcement of GDPR in the European Union and similar measures in other countries, protecting PII (Personally Identifiable Information) became a necessary step in preparing datasets or any processing of personal information. A person’s face, of course, is one the most prominent features of a person’s identity and thus there are many solutions that can detect and anonymize faces. However, they are hard to customize and automate. This is a DIY guide for those who want to do it by themselves using open source. The entire code sample can be run in Colab and is available here.

The steps involved in automatically anonymizing faces are:

  1. Detect faces using a pre-trained Deep Neural…

--

--

Changsin Lee
Changsin Lee

Written by Changsin Lee

AI/ML Enthusiast | Software Engineer | ex-Microsoftie | ex-Amazonian

No responses yet